International Journal of Mining Science and Technology 30 (2020) 179–188

Contents lists available at ScienceDirect

International Journal of Mining Science and Technology

journal homepage: www.elsevier.com/locate/ijmst

Use of drilling performance to improve rock-breakage efficiencies: A part

of mine-to-mill optimization studies in a hard-rock mine
Junhyeok Park, Kwangmin Kim ⇑
Department of Mining and Geological Engineering, University of Arizona, Tucson, AZ 85721, USA

a r t i c l e i n f o a b s t r a c t

Article history: In a hard-rock mine, blasting is an important rock-breakage process that impacts energy consumption
Received 31 March 2019 both in downstream comminution processes and mine productivity. Optimizing the blast fragmentation
Received in revised form 20 August 2019 to improve rock-breakage efficiencies during crushing and grinding is key to mine-to-mill (MTM) opti-
Accepted 20 December 2019
mization. This study explores the use of monitoring while drilling (MWD) data to achieve this goal.
Available online 30 December 2019
Representative penetration rates (PRs) were extracted from blastholes to estimate intact rock properties
and predict the breakage efficiencies that directly affect comminution energy consumption. Two intact
Keywords:
rock properties, tensile strength (TS) and Bond work index (BWI), were correlated with the PR data to pre-
Monitoring while drilling (MWD)
Mine-to-mill (MTM) optimization
dict these efficiencies in crushing and grinding, respectively. Because of the complexity of the raw MWD
Rock-breakage efficiency data and effects of various disturbances, the MWD data was preprocessed and normalized to achieve a
Drilling penetration rates representative PR value at each blasthole. This preprocessing entailed defining valid PR ranges from
the MWD data that could eliminate the noise related to discontinuity features in the rock mass structure
as well as errors in operator behaviors. The PR data was also normalized using the adjusted penetration
rate (APR) to minimize the effects of mechanical factors such as drill feed force, torque, and rotational
speed. To correlate the representative APR value with intact rock properties, TS and BWI, various labora-
tory experiments were conducted: drilling tests using a high-precision coring machine, Brazilian disc
tests, and Bond grindability tests. Based on the results of these experiments, models were developed
to predict rock-breakage efficiencies during crushing and grinding based on APR. The result of this study
can be used to obtain blast energy designs that consider comminution energy consumption and efficiency
in the downstream rock-breakage processes.
Ó 2020 Published by Elsevier B.V. on behalf of China University of Mining & Technology. This is an open
access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction
Crushing and grinding are some of the most energy-intensive
industrial processes in the world. In mines, rock-breakage (com-
minution) processes comprise a major portion–about 30%-60%–of
the total energy consumption [1–5]. These processes impact not
only energy consumption but also mine productivity. Over the last
few decades, efforts have been made to improve the comminution
efficiency at hard-rock mines, and one of the most popular of these
approaches is mine-to-mill (MTM) optimization, which was first
introduced by Julius Kruttschnitt Mineral Research Centre. The
ultimate goal of MTM is to improve the energy efficiency of rock
breakage by optimizing blast fragmentation. Blasting generates
microcracks that reduce the required comminution energy in
crushing and grinding, resulting in productivity increases [6,7].
⇑ Corresponding author.
E-mail address: kimkm@email.arizona.edu (K. Kim).
MTM research falls under two main categories: determining the
optimal (target) blast fragmentation and controlling the blast frag-
mentation. This study falls under the former category and consid-
ers the effect on downstream comminution processes, mainly
crushing and grinding, to optimize the blast fragmentation.
The MTM process for optimizing the blast fragmentation (Fig. 1)
is complex and must consider rock mass characteristics, the blast
energy, and downstream comminution processes. Rock mass char-
acteristics are especially important but challenging to determine in
real time since acquiring this information is labor-intensive and
time-consuming; these characteristics are typically measured
directly from laboratory and field tests. Furthermore, approaches
for applying rock mass characteristics in blast energy design are
limited; in practice, it is almost impossible to represent the entire
rock mass in real time given the scale and frequency of blasting
and the highly varied rock properties typically found at hard-
rock mines. In a porphyry copper deposit, for example, rock prop-
erties usually vary dramatically over even small domains because
of the complexity of the alteration zone and its interaction with

https://doi.org/10.1016/j.ijmst.2019.12.021
2095-2686/Ó 2020 Published by Elsevier B.V. on behalf of China University of Mining & Technology.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
180 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188
Fig. 1. Concept of MTM.
hydrothermal fluid [8]. In addition, access to the muckpile or pit
highwall to obtain samples may prompt safety concerns.
Drilling performance data can provide a viable means of charac-
terizing the rock mass for MTM. The use of blasthole drilling data is
advantageous since it is acquired systematically, routinely, and in
real time [9]. In the mining industry, monitoring while drilling
(MWD) is a standard practice for obtaining penetration rate (PR),
rotational speed, flushing pressure, and GPS-based drillhole posi-
tioning data [10]. Various data logging systems are commercially
available for recording this information, including terrain, drilling
efficiency indicator (DEI)TM, pro-visionTM, and the rig control system
(RCS)TM. Many researchers have tried using drilling performance
data to characterize rock properties and rock mass conditions
[11–13]. Segui and Higgins identified the rock hardness from
MWD data for use in a blast design [10]. Ghosh et al. developed
a contour map of rock mass strength based on the measured
MWD data in the Aitik mine [14]. Chen and Yue applied MWD data
to understand rock weathering grades [15]. Hatherly et al. verified
the validity of MWD data for classifying rock types by comparing it
with geophysical logging data from a drillhole probe [16]. And
MWD data was also used to detect geological boundaries such as
coal seams [17,18]. Recently, machine learning technology has
been applied to classify rock types using MWD data [19–21]. How-
ever, the current MWD technologies have still not been fully devel-
oped to accurately delineate and understand rock mass conditions,
especially for MTM.
MWD data includes two main types of information about the
rock mass condition: its intact rock properties and its structural
Fig. 2. Various sensing technologies and MWD systems for rock characterization and MTM in a hard-rock mine.
features (discontinuities). This study considers each separately
and focuses on intact rock properties, which can provide a mea-
sure of the rock-breakage efficiency during the crushing and
grinding processes and, ultimately, of the target blast fragmenta-
tion. This approach could simplify and reduce the high level of
uncertainty associated with the field data. Various state-of-the-
art remote sensing technologies are available for observing struc-
tural features and can be used with the MWD system described
below to investigate rock mass conditions in a mine. These tech-
nologies include LiDAR and photogrammetry instrumentation
that can be mounted on unmanned aerial vehicles (UAVs), as
shown in Fig. 2.
This study reasonably assumes that intact rock properties con-
trol the breakage efficiencies in the crushing and grinding pro-
cesses. For blast fragmentation control, rock mass structures
must be considered in addition to intact rock properties [9]. How-
ever, once the rock is blasted, its breakage efficiency depends
mainly on its intact properties since the particle sizes in the crush-
ing and grinding processes are much smaller than the typical block
size in rock mass [22]. Two parameters can be used to represent
intact rock properties for predicting the comminution efficiency:
tensile strength (TS) and Bond work index (BWI). TS is fundamen-
tal to the crack-opening stages of crushing and BWI is an intrinsic
material characteristic for the fine grinding process [23,24]. These
two parameters are inversely proportional to the energy efficien-
cies for crushing and grinding; higher TS and BWI require more
energy, which results in lower breakage efficiencies.
2. Preprocessing of drilling performance (MWD) data
MWD systems record the real-time drilling performance data
using sensors. Field MWD data usually includes some errors and
inconsistent PR data that may originate from several sources: the
drilling equipment itself, the media (due to geological variations),
and operator behaviors. For example, PR fluctuates when the drill
bit contacts discontinuities such as faults, joints, fractures, and
bedding planes, and operators may stop drilling to deal with unex-
pected situations such as rod or shift changes. Given the frequency
and scale of blasting in hard-rock mines, a significant amount of
MWD data must be updated daily. Therefore, the errors and incon-
sistencies in the PR data have a significant effect on the ability to
achieve optimum blast fragmentation. Consequently, it is impor-
tant to simplify the PR data, minimize these errors, and obtain a
consistent PR within a rock type to use drilling performance as a
measure of intact rock properties for MTM.
 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188
Fig. 3 shows the analysis of approximately 986,000 PR measure-
ments at a hard-rock mine obtained using a Pit Viper RCS system.
The data distribution was positively skewed, meaning that the
dataset contained a large amount of exceptional values; therefore,
simple averaging could produce misleading results. To simplify the
PR data and minimize errors, values higher than 1 m/min (about
5.4% of total population) were considered outliers and treated as
exceptional and invalid since including them would result in
unreasonably high PR averages. For example, the average of the
collected PR data was 0.4 m/min, but the median was almost half,
0.22 m/min, and the median was still much higher than the most
frequent value (mode), 0.14 m/min. These values illustrate the
complexity of conducting statistical analyses for MWD data. It is
notable that the 1 m/min threshold was based on the overall range
of PR data in the mine as well as interviews with drill operators;
consequently, this value is site specific and can vary in each mine.
2.1. Defining a valid PR range
To address these issues, a valid PR range was defined to repre-
sent intact rock properties. Data was collected from a single blast-
hole (around 15 m deep) in a hard-rock mine, as shown in Fig. 4.
Fig. 3. Histogram of PR data from field measurement.
Fig. 4. Example of valid data points from raw PR data.
181
Even though the geological domain was relatively consistent
throughout the borehole, some data points did attain exceptional
PR values (over 2 m/min) and fluctuated widely due to operational
errors and geological features. For example, the wide range of PR
values observed in the first 5 min was not considered valid. In
addition to operational errors, these values may represent struc-
tural features such as voids and fractures as well as mixed intact
rock properties. Since the rock type was consistent throughout
the entire hole, it was reasonably assumed that the PR range could
represent the intact properties for it was relatively consistent,
steady, and below the 1 m/min threshold (Fig. 4).
2.2. Normalization using APR
Although a PR range can be defined to minimize the distur-
bances related to geological variations and operator behaviors, it
does not apply to mechanical factors such as feed force, torque,
and rotational speed. To estimate intact rock properties using dril-
ling performance for MTM, the dependency of raw PR on mechan-
ical factors must also be minimized. Many researchers have tried to
find a function to obtain a representative index for estimating geo-
logical conditions from raw PR data; however, their results are
challengeable because the theoretical approach to describing
drill-rock interactions is limited to account for external distur-
bances [9]. Two popular approaches have been used to infer the
rock strength from MWD data: applying the specific energy of dril-
ling (SED) based on empirical analysis and applying the APR for
data normalization.
The SED for rotary non-percussion drilling is defined as the
work done to eliminate a unit volume of rock [25]. It can be calcu-
lated as:


4 2pxs
SED ¼ Fn þ ð1Þ
pD 2 v
where SED is the specific energy of drilling; D the diameter of the
drill; F n the feed force; x the rotational speed; s the torque; and
v the penetration rate.
SED is the most popular approach due to its simplicity; it infers
the rock strength index simply by adding the measured drilling
parameters into Eq. (1) [26]. However, SED is based on empirical
analysis and has no normalization process [9]. Consequently, it
has a limited ability to eliminate the effect of the mechanical fac-
tors and limited applicability for measuring intact rock properties
for MTM. For a rotary drill, the APR was suggested to be applied
in this study [20]. The effect of the mechanical factors was mini-
mized by normalizing the PR data using APR, as calculated by Eq.
(2) [18]. It is notable that, although this study focuses on a rotary
drilling, this approach could be applied to other drilling systems,
pending further studies.
v
APR ¼ a pffiffiffiffiffiffiffi ð2Þ
F n xs
pffiffiffiffiffiffiffiffiffiffi
a ¼ F n;r xr sr ð3Þ
where the coefficient a is determined by reference values, which are
represented by the subscript ‘‘r”.
The reference values consider the rock type, rock strength, dril-
ling method, and operational conditions. To determine these val-
ues, the maximum change ranges must first be defined for three
drill variables–feed force, rotational speed, and torque–while con-
sidering the possible operational conditions in a mine. The most
representative (median) values can then be assigned as reference
values using the results of baseline tests that entail applying the
three controllable drill variables within the defined ranges. In this
study, an electric motor drill was used. Since the drill’s power out-
put is constant and the rotational speed and torque of its motor are
182 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188
inversely proportional, the variable for torque can be ignored in Eq.
(2). However, in a drill powered by an internal combustion engine
(the standard in mines), feed force, rotational speed, and torque
respond independently, and torque (s) varies depending on the
revolutions per minute (RPM) and power of the engine [27]. The
reference values in this study were defined as 430 RPM (rotational
speed) and 758 KPa (feed force). It is notable that the same values
Fig. 5. Modified GCTS pressure-controlled coring machine (RCD-250) for a drilling
test.
Fig. 6. Results of normalizing the laboratory drill data.
were used for sandstone and limestone since the difference in
strength between these rock types was insignificant in this study
and the tests were conducted in well-controlled environments
using an electric drill.
The APR was calculated using Eqs. (2) and (3) with these
defined reference values. For example, if a PR is measured with
420 RPM and 800 KPa of feed force, the denominator becomes
pffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffi
800 420s. The value of acan be expressed as 758 430s and
divided by denominator. As mentioned above, torque (s) is con-
stant since an electric motor drill was used. With torque (s) can-
celed, the APR turns out to be about 96% of the PR value.
A modified GCTS pressure-controlled coring machine (RCD-250)
was used for the drilling tests with 24.5 mm core bits (9 mm dia-
mond segment height, 3 mm segment thickness, and 102 mm
overall height). The drilling depth, rotational speed, and feed force
of the bit were measured using a real-time multi-channel data log-
ger. Typical drill progress ranges were between 25 and 40 mm,
depending on the tested sample lengths available. Fig. 5 shows
the coring machine and set of sensors for the laboratory-scale
MWD test.
The disturbances generated from the mechanical factors were
successfully eliminated (normalized) using the APR, and consistent
PR data was obtained. Fig. 6 shows the results of normalizing the
laboratory drill data: the raw PR data (Fig. 6a), which were highly
dependent on feed force and rotational speed changes, and the PR
data after applying APR (Fig. 6b).
Fig. 7 shows histograms constructed using PR and APR data that
falls within the valid range to compare their overall distributions.
Although the trends for the two histograms do not differ signifi-
cantly, the APR data is more distributed and less centralized than
the PR data.
The results shown in Figs. 6 and 7 are based on laboratory PR
test data; applying them to field conditions is premature. Field
applications will require further studies–in particular, studies fea-
turing drills that use internal combustion engines. For example, the
APR equation (Eq. (2)) must consider a specific torque curve that
can be provided by equipment manufacturers.
3. APR and intact rock properties
Processed MWD data can be used to predict the rock-breakage
efficiencies in crushing and grinding processes based on the rela-
tionship between drilling performance (APR) and intact rock prop-
erties. A series of laboratory rock tests–drilling tests using a high-
precision coring machine, tensile tests, and BWI tests–were con-
ducted to find the relationship between drilling performance and
intact rock properties. As previously noted, this study assumed that
 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188 183

Fig. 7. Comparisons of distributions of PR and APR data that falls within the valid range.

Fig. 8. Rock samples and laboratory experiment procedures.

TS and BWI could represent the resistance to fracturing and abra- circulation load remained at 250% for three tumbling periods, it
sion during the comminution; TS can be considered as a funda- was regarded as reaching a steady state.
mental property for the rock breakage in crushing and BWI is the Based on test results, BWI was calculated using Eq. (4). This cal-
most popular measure of the energy required in fine-grinding [27]. culation used product/feed particle sizes along with the average
In the laboratory tests, the tested rock types included Coconino net grams produced per revolution (Gpr) measured during the last
sandstone as well as limestone from a quarry in northern Arizona. three periods after reaching a steady state. P80 and F 80 were
A total of 141 samples (80 for the sandstone, 61 for the limestone) obtained from the Gaudin-Schumann distribution curve plotted
was used for the tests. The NX core (54.7 mm diameter) with by sieving tests using various mesh sizes.
100 mm of height was divided into two parts: 25 mm for Brazilian
1:1  44:5
disc tests (to measure TS) and 75 mm for drilling tests (to measure BWI ¼
 ð4Þ
APR). Fragments of the tested samples were then collected, Pi 0:23
 Gpr0:82  p10ffiffiffiffiffi  p10ffiffiffiffiffi
P80 F 80
crushed using a laboratory-scale jaw crusher, and used for the
grindability tests to determine the BWI. It is notable that fewer
Bond grindability tests were conducted because they require a lar-
ger amount of feed samples (over 5 kg) than the Brazilian disc and
drilling tests. Fig. 8 shows the tested rock samples and the proce-
dure for the laboratory experiments.

3.1. Grindability tests

Bond ball mill tests were conducted to measure the grindability

of rock. BWI is an intrinsic characteristic of rock grindability, rep-
resenting the specific energy for grinding rock when the feed size
is 6 mesh and the product size is 100 mesh. This study used a lab-
oratory Bond ball mill consisting of a cylindrical steel jar and a 70
RPM motor to derive BWI. Fig. 9 shows the mill.
To measure BWI, the standard Bond grindability test procedure
was followed, as shown in Fig. 10 [28,29]. This test requires feed
material (rock particles) of less than 6 mesh (3350 lm). These par-
ticles were obtained by crushing the aftermath fragments from the
drilling and Brazilian disc tests with the laboratory jaw crusher.
Each tumbling procedure was repeated until the material flow
reached the steady state of the circulation load, which was calcu-
lated by using the ratio of the weight of oversize to undersize frag-
ments based on the 100 mesh (150 lm) sieve opening. Once the Fig. 9. Laboratory Bond ball mill.
184 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188

Fig. 10. Procedure of the Bond standard grindability test.

where BWI is the Bond work index; Pi the opening sieve size; Gpr
the grams produced per revolution; P80 80% passing size of product;
and F 80 80% passing size of feed.
For limestone samples, Eq. (4) was applicable, but for the sand-
stone samples, a different approach was used to calculate BWI.
Since sandstone is a typical clastic sedimentary rock consisting of
100–200 lm silicate particles, the cohesion between particles is
much weaker than their strength. This characteristic caused an
inconsistent phase of the grinding speed in the tumbling mill.
The particle size quickly reduced to 100–200 lm, but additional
grinding at finer sizes produced almost no change due to the high
hardness of the silicate. Therefore, it was difficult to achieve a
steady state for calculating Gpr. As an alternative, the Berry and
Bruce comparative method was used to estimate the BWI [30].
The applicability of this method has been verified to measure the
BWI values of graphite and columbite ores, respectively, and
showed that their values were within the range of the actual
BWI values of the ores [31,32]. Likewise, it was found that the
BWI estimates from this comparative method were within 8% of
the actual BWI [33]. Given the purpose of this study–identifying
the relative changes of BWI in accordance with the blasthole APR
changes and predicting/improving the rock-breakage efficiency–
this method was acceptable. Although it does not yield an actual
BWI, it provides consistent test results, which is key to this study.
The BWIs of sandstones were approximated using Eq. (5):


p10
ffiffiffiffiffiffi  p10
ffiffiffiffiffiffi
P 80r F 80r
BWIt ¼ BWIr 
 ð5Þ
p10
ffiffiffiffiffiffi  p10
ffiffiffiffiffiffi
P 80t F 80t
where r and t refers to the referenced and tested material,
respectively.
The sandstone sample that was chosen as a reference material
(sandstone No. 1 in Table 1) had a more cohesive structure than
other specimens and did not exhibit the general propensity of
sandstone as described above.
The average BWIs for sandstone and limestone were 5.84 ± 1.51
and 6.45 ± 0.61 kWh/t, respectively. The measured BWIs were still
considered acceptable given the purpose of this study and the vari-
ability of rock, although the BWIs were lower than the general
ranges for sandstone (11–38 kWh/t) and limestone (4–26 kWh/t)
[34].
Table 1 shows details for the tested rock samples as well as the
experiment results. The Leeb hardness (LH) index was measured to
investigate the basic strength of the rock samples using the
method proposed by Aoki and Matsukura [35]. The unconfined
compressive strength (UCS) was estimated by converting the LH
index [36]. Because the number of rock samples was limited, other
basic mechanical properties for the rock types were not evaluated
in this study, and only one BWI measurement was conducted for
each test set.
3.2. Prediction models
Models were established to predict rock-breakage efficiencies
using nonlinear regression analysis. BWI and TS can be predicted
using the measured APR. Previous studies have shown that the
relationship between PR and the rock strength commonly follows
a power or exponential function [37–40]. The test results in this

Table 1
Information for tested rock samples and experiment results.

Rock type No. Number of Specific LH index UCS (MPa) APR (mean ± SD) TS (mean ± SD) BWI
samples gravity (mean ± SD) [36] (cm/min) (MPa) (KWh/t)
Sandstone 1 10 2.58 692.20 ± 21.68 104.84 ± 2.86 4.11 ± 1.60 15.64 ± 1.98 9.66
2 15 2.33 618.80 ± 17.48 70.69 ± 2.80 10.02 ± 0.66 5.59 ± 0.68 5.47
3 10 2.16 552.07 ± 17.34 49.40 ± 2.80 7.48 ± 0.32 5.44 ± 0.96 5.56
4 10 2.27 597.43 ± 25.46 63.02 ± 2.92 9.01 ± 1.68 5.63 ± 0.95 4.94
5 10 2.33 524.40 ± 19.42 42.58 ± 2.83 13.43 ± 0.42 5.99 ± 0.54 4.90
6 6 2.34 767.12 ± 17.81 156.76 ± 2.80 5.30 ± 0.39 6.81 ± 0.85 7.07
7 8 2.08 466.13 ± 33.22 31.14 ± 3.05 12.69 ± 1.10 3.93 ± 0.43 4.87
8 5 2.39 631.65 ± 16.28 75.74 ± 2.78 7.34 ± 0.71 5.41 ± 0.47 4.66
9 6 2.33 527.45 ± 14.94 43.28 ± 2.76 13.71 ± 3.01 3.33 ± 0.46 5.45
Limestone 1 4 2.64 568.87 ± 17.17 54.06 ± 2.79 4.75 ± 0.29 5.15 ± 0.44 5.73
2 4 2.63 634.30 ± 10.14 76.82 ± 2.69 4.81 ± 0.13 5.89 ± 0.75 6.68
3 9 2.64 570.57 ± 13.27 54.56 ± 2.74 4.74 ± 0.40 5.36 ± 0.58 6.74
4 9 2.64 578.67 ± 19.85 56.98 ± 2.83 4.77 ± 0.3 6.10 ± 0.75 7.11
5 7 2.62 583.87 ± 15.65 58.60 ± 2.77 4.91 ± 0.21 5.02 ± 0.97 6.58
6 4 2.63 608.53 ± 10.28 66.89 ± 2.69 4.91 ± 0.55 5.05 ± 1.03 5.20
7 6 2.67 572.20 ± 19.04 55.04 ± 2.82 4.27 ± 0.37 6.76 ± 0.95 6.80
8 3 2.66 578.33 ± 11.46 56.88 ± 2.71 4.42 ± 0.10 5.27 ± 1.03 6.76
9 10 2.64 630.43 ± 10.74 75.24 ± 2.70 4.76 ± 0.47 5.21 ± 0.81 6.06
10 5 2.67 587.20 ± 14.21 59.65 ± 2.75 3.97 ± 0.21 6.72 ± 0.38 7.29
 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188
study followed a power function (Y ¼ aX k ) that can be expressed as
a linear function in a log-log domain. Applying logarithmic trans-
formations in a regression model is a common way to handle a
nonlinear relationship between two variables, especially when
they are highly skewed like the field MWD data, as shown in Fig. 3.
Fig. 11 shows the test results for the 80 sandstone samples. The
model plots these results with a 95% confidence interval (CI) and
prediction interval (PI) lines. It is notable that the CI indicates
the 95% probability range in which a current observation lies, while
PI indicates the range in which a future observation will fall with
95% probability.
The upper and lower PI boundaries were defined for the predic-
tion model and the relationship between APR and TS, with the non-
linear plot (Fig. 11a) and the log-log plot (Fig. 11b). The prediction
model for TS was generated using Eq. (6):
log TS ¼ 0:74log APR þ 1:46
where APR is the adjusted penetration rate; and TS the tensile
strength.
For APR values lower than 5 cm/min, the corresponding TS
increased exponentially, indicating that the energy required for
crushing is exponentially higher in this range (Fig. 11a). For exam-
ple, when APR was 5 cm/min, the predicted TS was 8.8 MPa. When
APR decreased from 8 to 5 cm/min (a change of 3 cm/min), the TS
increase was only 2.6 MPa; however, when it decreased from 5 to
2 cm/min (the same change), TS increased by 8.5 MPa. Conse-
quently, if the measured APR in the field decreased, more energy
Fig. 11. Regression model of sandstone for APR versus TS.
Fig. 12. Regression model of sandstone for APR versus BWI.
185
will be required in the crushing process after blasting, but the
required blast design to improve the rock-breakage efficiency will
differ significantly depending on the APR range. APRs that are
lower than 5 cm/min will require more blasting energy if they
decrease based on the relationship shown in Fig. 11.
Fig. 12 plots BWI as a function of APR for sandstone. Data points
were limited because of a shortage in feed materials (sandstone
samples). Since BWI experiments require a relatively large amount
of samples, more tests are required to generate the prediction
model; however, this analysis did show that BWI relied signifi-
cantly on APR changes.
Based on the results shown in Fig. 12, a preliminary regression
(prediction) model of APR versus BWI was generated for the sand-
stone, as shown in Eq. (7):
log BWI ¼ 0:43log APR þ 1:16 ð7Þ
ð6Þ
For APRs lower than around 7 cm/min, the corresponding BWI
increased exponentially with decreasing APR (harder rock for
grinding) (Fig. 12a).
Figs. 13 and 14 show the test results for the limestone samples,
which vary less than the sandstone results. This difference could be
related to the sample consistency (the limestone was sampled at a
single quarry). More studies are required to determine the cause of
this difference.
Figs. 13 and 14 showed a clear dependency of TS and BWI on
APR changes, but the data points were dispersed and too few in
number to develop reliable prediction models. Since the goal of
186 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188
Fig. 13. Regression model of limestone for APR versus TS.
Fig. 14. Regression model of limestone for APR versus BWI.
this study is to demonstrate the use of PR data in mining applica-
tions, preliminary regression models of APR versus TS/BWI were
generated for limestone as shown in Eqs. (8) and (9).
log TS ¼ 0:96log APR þ 1:38
log BWI ¼ 0:83log APR þ 1:38
For APRs below around 4 cm/min, the corresponding TS and
BWI increased exponentially with decreasing APR (Figs. 13a and
14a).
This analysis found that rock-breakage efficiencies (TS and BWI)
could be predicted using APR and that the prediction models fol-
lowed a power function (Y ¼ aX k ). Because the prediction model
was both site-type-specific and rock-type-specific, it was possible
to minimize the variation that may occur even within a single rock
type in a mine. Therefore, the same specificity applies to the coef-
ficients of the power function (‘‘a” and ‘‘k”). The coefficients can be
derived from laboratory tests initially and later updated to mini-
mize the discrepancy between the predicted and monitored values
based on the data obtained during blasting, crushing and grinding.
Through this iterative process, reliable prediction models can even-
tually be developed for a mine.
4. Potential field applications
For practical application in the field, the results of this analysis
need to be simplified and presented in a way that is useful for mine
operators. This could be accomplished by developing thematic
maps for various blasthole classifications (e.g. ‘‘soft”, ‘‘medium”,
and ‘‘hard”) that could serve as a guide before blasting, especially
ð8Þ in the absence of geological surveys.
This would entail developing a real-time thematic map that
ð9Þ shows rock-breakage characteristics (crushability and grindability)
based on representative APR values (MWD data) and prediction
models. Such a map could be created by integrating MWD data into
mine planning software such as MineSightTM, SurpacTM, VulcanTM,
etc.; the data could be simply integrated into the existing blasthole
database, which allows users to store, manage and visualize data.
Fig. 15 describes this concept. It is notable that applying this
method in the field still poses many challenges and requires fur-
ther study; the purpose of this section is to illustrate the concept
of the potential applications in mining industry.
APR map displaying the representative APR value at each blast-
hole can be generated, as shown on the left side of Fig. 15.
Although each blasthole has a representative APR value, in prac-
tice, it is almost impossible for mine operators to apply different
blast energies at each hole [10]. Therefore, this map could be sim-
plified by classifying blastholes–for example, as ‘‘soft”, ‘‘medium”,
and ‘‘hard”. It is notable that this classification refers to the rock-
breakage efficiency such as crushability and grindability based on
TS and BWI–not rock hardness. For example, two thresholds, 5
and 10 cm/min, could be defined in Fig. 11, which show the rela-
tionship between APR and TS. Blastholes having APR values that
are lower than 5 cm/min could be classified as ‘‘hard”, between 5
 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188
Fig. 15. Thematic mapping using drilling performance.
and 10 cm/min as ‘‘medium”, and higher than 10 cm/min as ‘‘soft”.
At each classification, a blast energy design could be assigned to
improve the rock-breakage efficiency in the crushing process.
Two simplified thematic maps could be generated based on the
prediction models, as shown in Fig. 15: a TS map for presenting
the comminution efficiency in the crushing process and a BWI
map for presenting the comminution efficiency in the grinding
process.
An optimal blast energy design requires both determining the
target fragmentation and controlling the blast fragmentation. This
study focused only on the target blast fragmentation, which must
consider downstream comminution processes (crushing and grind-
ing). Therefore, the use of thematic maps as suggested above is
limited to identifying the target fragmentation; harder rock for
crushing and grinding demands better (smaller) blast fragmenta-
tion, which in turn requires higher blast energy to improve the
rock-breakage efficiency and reduce the downstream comminu-
tion energy. Blast fragmentation control, on the other hand, must
consider factors such as rock mass strength and structural features
that were not part of this study. Further studies are required to
finalize the optimal blast energy (powder factor) design for a given
rock mass.
5. Conclusions and future work
This study showed the potential for using drilling performance
data to estimate rock-breakage characteristics in crushing and
grinding as the part MTM efforts. A valid PR range was defined to
minimize the effects of geological factors and represent intact rock
properties–TS and BWI–from blasthole PR data. Likewise, the APR
concept was applied to minimize the effect of mechanical factors
by normalizing the PR data. The results of laboratory tests with
sandstone and limestone samples indicated that APR could be cor-
related with TS and BWI, which were directly related to rock-
breakage efficiencies in the crushing and grinding processes,
respectively. Site-type-specific and rock-type-specific models were
then generated to predict rock-breakage efficiencies with a repre-
sentative APR at each blasthole. In addition, a methodology was
proposed for field applications that involves creating thematic
maps (possibly using mine planning software) to guide operators
in allocating the proper blast energy design at blastholes. To create
these maps, blastholes would be classified based on their rock-
breakage characteristics (e.g. ‘‘soft”, ‘‘medium”, or ‘‘hard”). This
study demonstrated how to extract intact rock properties for pre-
dicting rock-breakage characteristics in the crushing and grinding
processes from drilling performance (MWD) data. It also showed
how to develop prediction models for estimating rock-breakage
characteristics using APR values from laboratory drilling, Brazilian
disc, and Bond grindability tests.
187
Although raw MWD field data were used to define the valid PR
ranges for extracting the intact rock properties, the prediction
models were generated based on the results of laboratory tests
only. Some challenges must be overcome before this methodology
can be applied in the field. One such challenge is to understand the
heterogeneous nature of the rock mass and effectively extract the
intact rock properties. Although this study extracted valid PR
ranges and obtained APR values via simple algorithms, this was
achieved under fully controlled conditions in the laboratory; actual
field data may require much more sophisticated and complex sta-
tistical methods. Another challenge is to correlate the field and lab-
oratory APR values. Obtaining representative APR values from raw
field MWD data must account for various conditions such as signif-
icant interactions between the drill bit and rock, friction between
the drill rods and bore wall, the use of drilling fluid, and bit wear
[16]. Therefore, developing a reliable correlation between labora-
tory APR and field APR currently involves a significant uncertainty.
Further follow-up studies will be conducted to achieve reliable
field APR values, site-specific prediction models and, ultimately,
site-specific target blast fragmentation that considers the down-
stream comminution processes in a hard-rock mine.
References
[1] Powell M, Bye A. Beyond mine-to-mill: Circuit design for energy efficient
resource utilization. In: Proceedings of the 10th mill operators’
conference. Adelaide: Australasian Institute of Mining and Metallurgy
(AusIMM); 2009. p. 357–64.
[2] Ballantyne GR, Powell MS, Tiang M. Proportion of energy attributable to
comminution. In: Proceedings of the 11th mill operators’
conference. Hobart: Australasian Institute of Mining and Metallurgy
(AusIMM); 2012. p. 25–30.
[3] Musa F, Morrison R. A more sustainable approach to assessing comminution
efficiency. Miner Eng 2009;22(7–8):593–601.
[4] Norgate T, Haque N. Energy and greenhouse gas impacts of mining and mineral
processing operations. J Cleaner Prod 2010;18(3):266–74.
[5] Napier-Munn T. Is progress in energy-efficient comminution doomed?. Miner
Eng 2015;73:1–6.
[6] Nielsen K, Malvik T. Grindability enhancement by blast-induced microcracks.
Powder Technol 1999;105(1–3):52–6.
[7] Michaux S, Djordjevic N. Influence of explosive energy on the strength of the
rock fragments and SAG mill throughput. Miner Eng 2005;18(4):439–48.
[8] Bodnar RJ, Beane RE. Temporal and spatial variations in hydrothermal fluid
characteristics during vein filling in pre ore cover overlying deeply buried
porphyry copper-type mineralization at Red Mountain, Arizona. Econ Geol
1980;75(6):876–93.
[9] Rai P, Schunesson H, Lindqvist PA, Kumar U. An overview on measurement-
while-drilling technique and its scope in excavation industry. J Instit Eng
(India): Series D 2015;96(1):57–66.
[10] Segui JB, Higgins M. Blast design using measurement while drilling
parameters. Fragblast 2002;6(3–4):287–99.
[11] Scoble MJ, Peck J. A technique for ground characterization using automated
production drill monitoring. Int J Surf Min Reclam Environ 1987;1(1):41–54.
[12] Vynne J. The application and economic benefits of blasthole drill monitors. In:
Proceedings of the annual conference on explosives and blasting
technique. Nevada: International Society of Explosives Engineers; 1997. p.
635–46.
188 J. Park, K. Kim / International Journal of Mining Science and Technology 30 (2020) 179–188
[13] Kahraman S, Rostami J, Naeimipour A. Review of ground characterization by
using instrumented drills for underground mining and construction. Rock
Mech Rock Eng 2016;49(2):585–602.
[14] Ghosh R, Schunnesson H, Kumar U. Evaluation of rock mass characteristics
using measurement while drilling in boliden minerals aitik copper mine,
Sweden. In: Proceedings of the 22nd MPES conference. Dresden: Mine
Planning and Equipment Selection; 2014. p. 81–91.
[15] Chen J, Yue ZQ. Ground characterization using breaking-action-based zoning
analysis of rotary-percussive instrumented drilling. Int J Rock Mech Min Sci
2015;75:33–43.
[16] Hatherly P, Leung R, Scheding S, Robinson D. Drill monitoring results reveal
geological conditions in blasthole drilling. Int J Rock Mech Min Sci 2015;100
(78):144–54.
[17] Ahsan N, Scheding S, Monteiro ST, Leung R, McHugh C, Robinson D. Adaptive
sampling applied to blast-hole drilling in surface mining. Int J Rock Mech Min
Sci 2015;75:244–55.
[18] Leung R, Scheding S. Automated coal seam detection using a modulated
specific energy measure in a monitor-while-drilling context. Int J Rock Mech
Min Sci 2015;75:196–209.
[19] Kadkhodaie-Ilkhchi A, Monteiro ST, Ramos F, Hatherly P. Rock recognition
from MWD data: a comparative study of boosting, neural networks, and fuzzy
logic. IEEE Geosci Remote Sens Lett 2010;7(4):680–4.
[20] Zhou H, Hatherly P, Monteiro ST, Ramos F, Oppolzer F, Nettleton E, et al.
Automatic rock recognition from drilling performance data. In: Proceedings of
IEEE international conference on robotics and
automation. Minnesota: RiverCentre; 2012. p. 3407–12.
[21] Basarir H, Karpuz C. Preliminary estimation of rock mass strength using
diamond bit drilling operational parameters. Int J Min Reclam Environ 2016;30
(2):145–64.
[22] Isokangas E, Sonmez B, Wortley M, Valery W. Using SmartTag to track ore in
process integration and optimization projects: some case studies in a variety
of applications. The Southern African Institute of Mining and Metallurgy
Platinum; 2012. p. 871–92.
[23] Guo H, Aziz NI, Schmidt LC. Linear elastic crack tip modelling by the
displacement discontinuity method. Eng Fract Mech 1990;36(6):933–43.
[24] Sand GW, Subasinghe GKN. A novel approach to evaluating breakage
parameters and modelling batch grinding. Miner Eng 2004;17(11–12):1111–6.
[25] Teale R. The concept of specific energy in rock drilling. In: Proceedings of
international journal of rock mechanics and mining sciences & geomechanics
abstract. Pergamon; 1965. p. 57–73.
[26] Beattie N. Monitoring-while-drilling for open-pit mining in a hard rock
environment: an investigation of pattern recognition techniques applied to
rock identification Doctoral dissertation. Kingston: Queen’s University; 2009.
p. 66–90.
[27] Park J. Estimation of rock comminution characteristics by using drill
penetration rates Master’s dissertation. Arizona: The University of Arizona;
2016. p. 22–49.
[28] Levin J. Observations on the Bond standard grindability test, and a proposal for
a standard grindability test for fine materials. J South Afr Inst Min Metall
1989;89(1):13–21.
[29] Yap RF. Determination of the Bond work index using an ordinary laboratory
batch ball mill. Design and installation of comminution circuits; 1982, p. 176–
203..
[30] Berry TF, Bruce RW. A simple method of determining the grindability of ores.
Can Min J 1966;87:63–5.
[31] Adeoti MO, Dahunsi OA, Awopetu OO, Aramide FO, Alabi OO, Johnson OT, et al.
Determination of work index of graphite from Saman-Burkono (Nigeria) using
modified Bond’s method. Niger J Technol 2019;38(3):609–13.
[32] Alabi OO, Yaro SA, Dungka GT, Asuke F, Dauda ET. Determination of work index
of Gyel-Bukuru columbite ore in Plateau state, Nigeria. J Miner Mater Charact
Eng 2015;3:194–203.
[33] Todorovic D, Trumic M, Andric L, Milosevic V. A quick method for Bond work
index approximate value determination. Physicochem Probl Miner Process
2017;53(1):321–32.
[34] Darling PSME. mining engineering handbook. SME; 2011. p. 88.
[35] Aoki H, Matsukura Y. Estimating the unconfined compressive strength of
intact rocks from Equotip hardness. Bull Eng Geol Environ 2008;67(1):23–9.
[36] Asiri Y, Corkum A, El Naggar H. Leeb hardness test for UCS estimation of
sandstone. Canadian Geotechnical Society; 2016. p. 3–5.
[37] Bauer A, Calder PN. Open pit drilling-factors influencing drilling rates. In:
Proceedings of the 4th Canadian symposium of rock mechachics symposium,
CIMM annual meeting. Ottawa: Dept Energy; 1967. p. 1–8.
[38] Altindag R, Güney A. ISRM suggested method for determining the Shore
hardness value for rock. Int J Rock Mech Min Sci 2006;43(1):19–22.
[39] Servet D, Nazmi S, Ibrahim U, Tamer E, Deniz A, Rasit A. Variation of vertical
and horizontal drilling rates depending on some rock properties in the marble
quarries. Int J Min Sci Technol 2014;24(2):269–73.
[40] Ataei M, KaKaie R, Ghavidel M, Saeidi O. Drilling rate prediction of an open pit
mine using the rock mass drillability index. Int J Rock Mech Min Sci
2015;73:130–8.